Digital Twin-empowered Vehicular Edge Computing (DTVEC) integrates digital twin models with edge computing in vehicular networks to enable real-time, high-fidelity services such as autonomous driving and traffic management. These services depend on timely and temporally aligned data from multiple sources. To quantify this requirement, we introduce the Age of Fusible Information (AoFI), a novel metric that captures both data freshness and inter-source temporal alignment. Minimizing AoFI through distributed subchannel allocation across base stations (BSs) is a challenging NP-hard problem due to many-tomany relationships, inter-BS interference, and vehicle mobility. To address this, we propose a multi-agent automatic heuristic design (AHD) framework that leverages the interpretability and adaptability of large language models (LLMs) for dynamic, context-aware heuristic search. Unlike conventional learningbased methods, our approach discovers interpretable, resourceefficient heuristics guided by a centralized diagnostic signal called panorama. Extensive evaluations using real-world vehicle data show that LLM-generated heuristics outperform state-of-the-art learning-based and manually crafted baselines at minimizing AoFI and generalize well across diverse, unseen contexts.

Ensuring timely delivery is crucial with the increasing competition in online meal delivery services. This requires the industry to adopt new technologies and the corresponding operational models, including the use of drones. Concerning the desired features of meal delivery, such as safety and reliability, we propose an operational model that incorporates the usage of drones into the current rider-based delivery model. In our approach, known as drone resupply, drones transport meals from restaurants to riders, and riders then deliver them to customers. We aim to address two key issues when implementing this approach. First, at the operational level, models and algorithms are developed to effectively coordinate rider routing and drone scheduling. These algorithms are tailor-made by leveraging the short routes in the meal delivery industry. Second, at the tactical planning level, we reveal managerial insights to aid meal delivery platforms in making informed decisions regarding the implementation of drone resupply solutions. Particularly, drone resupply proves to be more efficient than rider-only mode across diverse order volumes and service ranges, and remains competitive when the promised delivery time is extended. The effectiveness of drone resupply is closely tied to the fleet configuration of riders and drones, as they have different yet complementary roles in achieving on-time delivery. Additionally, restricting one single order per drone trip does not compromise the effectiveness of drone resupply delivery, but necessitates more demanding drone schedules.

The dynamic stochastic multi-knapsack problem, which handles dynamically arrived requests, captures key features of real-time resource allocation in applications such as cloud computing and online advertising. Theoretically speaking, it is computationally challenging due to the strong NP-hardness inherited from the multi-knapsack structure and the curse of dimensionality arising from stochastic dynamics. In practice, some problems may reveal additional features that can facilitate problem solving. In this paper, we focus on such a case where requests belong to some item types. This enables us to develop a pattern-based control framework that explicitly exploits the combinatorial structure of packing decisions. Within this framework, we propose two complementary policies. The PB-BPC policy integrates packing patterns into dual pricing to obtain tighter upper bounds, and the PB-DPC policy further refines this approach by solving a pattern-level primal linear program via column generation and using the resulting allocation targets to guide online decisions. Moreover, we establish that the PB-DPC policy is asymptotically optimal. Its revenue loss relative to the deterministic linear programming upper bound is bounded by a constant independent of problem scale, implying that the relative optimality gap vanishes as the system size increases. Numerical experiments demonstrate the superiority of the proposed pattern-based policies across a variety of settings, achieving up to a 6\% improvement in revenue ratio over baseline methods in most cases. Our results provide a tractable and near-optimal solution framework for large-scale dynamic packing systems and yield actionable insights for resource allocation control in revenue management.

This article studies an appointment scheduling problem where a service provider dynamically receives appointment requests from a random number of customers. By leveraging the randomness of the number of potential customers, we develop a nonsequential appointment scheduling policy as an alternative to the conventional first-come-first-served (FCFS) policy. This allows for more flexibility in managing appointment scheduling. To calculate the optimal policy, we develop a branch-and-bound algorithm in which the lower bound is estimated using multiple approaches, such as optimality conditions, dynamic programming for calculating FCFS policy, and the shortest path reformulation. Through numerical studies, we observe that nonsequential appointment scheduling is particularly advantageous in systems characterized by highly fluctuating customer numbers or low congestion. In such cases, leaving gaps between appointments for potential future arrivals proves to be a more appropriate strategy. We also evaluate the performance of heuristics proposed in prior literature and provide insights into situations where these heuristics can be effectively applied.

For an unbalanced cooperative game, its grand coalition can be stabilized by some instruments, such as subsidization and penalization, that impose new cost terms to certain coalitions. In this paper, we study an alternative instrument, referred to as cost adjustment, that does not need to impose any new coalition-specific cost terms. Specifically, our approach is to adjust existing cost coefficients of the game under which (i) the game becomes balanced so that the grand coalition becomes stable, (ii) a desired way of cooperation is optimal for the grand coalition to adopt, and (iii) the total cost to be shared by the grand coalition is within a prescribed range. Focusing on a broad class of cooperative games, known as integer minimization games, we formulate the problem on how to optimize the cost adjustment as a constrained inverse optimization problem. We prove [Formula: see text]-hardness and derive easy-to-check feasibility conditions for the problem. Based on two linear programming reformulations, we develop two solution algorithms. One is a cutting-plane algorithm, which runs in polynomial time when the corresponding separation problem is polynomial time solvable. The other needs to explicitly derive all the inequalities of a linear program, which runs in polynomial time when the linear program contains only a polynomial number of inequalities. We apply our models and solution algorithms to two typical unbalanced games, including a weighted matching game and an uncapacitated facility location game, showing that their optimal cost adjustments can be obtained in polynomial time. History: Accepted by Area Editor Andrea Lodi for Design & Analysis of Algorithms—Discrete. Funding: This work was supported by the National Natural Science Foundation of China [Grants 72022018 and 72091210]; the Research Grants Council of the Hong Kong SAR, China [Grant 16210020]; Hong Kong Polytechnic University [Grant P0032007]; and the Youth Innovation Promotion Association, Chinese Academy of Sciences [Grant 2021454]. Supplemental Material: The online appendix is available at https://doi.org/10.1287/ijoc.2022.0268 .

In the service industry, a service provider may sell a collection of service activities as a package, also known as a service bundle. Empirical studies indicate that the customer's ex-post perception of a service bundle depends on not only the utility of each activity, but also the sequence of the activities being delivered. The latter can be measured by certain sequence effects, such as the utility of the peak activity, the utility of the end activity, and the trend of utility change over the activities. This phenomenon gives a service provider an opportunity to optimize a service bundle by manipulating the activities and their sequence. Such a service bundle design problem can be formulated as a 0-1 sum-of-ratios problem. To solve the problem, we design a novel geometric branch-and-bound algorithm. The algorithm divides the objective function into several dimensions, and repeatedly strengthens the bounds of each dimension. This enables us to convert the 0-1 sum-of-ratios problem into a series of 0-1 quadratic optimization problems. Computational experiments show that the algorithm can solve the service bundle design problem efficiently. (C) 2022 Elsevier B.V. All rights reserved.

This paper considers pricing and quality decisions for a product in a firm which is concerned about demand risk stemming from uncertainty as to customers' valuation. A stylized model is built on the basis of the expected utility theory, and the optimal solutions are characterized under very mild conditions in the scenarios of risk neutrality and risk aversion respectively. It is found that when either price or quality is the sole decision variable, the optimal price decreases with a higher level of risk aversion whereas the optimal quality increases with it. In comparison, when both price and quality are decision variables, a more risk-averse firm will set the product's price and quality at a lower level simultaneously. This is due to the mutual reinforcement effect between these two decision variables as strategic complements. This effect further enhances the impacts of risk aversion, leading to more radical changes in marginal profit and product price-performance as the risk aversion degree varies. Hence, compared to a single variable, risk aversion is a more serious issue in the presence of both the decision variables in question. (C) 2019 Elsevier Ltd. All rights reserved.

In the literature, different risk criteria are proposed for risk-averse decision-making, but it is often unclear which risk criteria should be used and what their differences may be. This study investigates this problem when a risk-averse, make-to-order firm decides its pricing and sales effort, two factors affecting demand. We consider alternative risk criteria commonly used in the literature, namely expected utility theory (EUT), mean-variance (MV), mean-semideviation (MS), value-at-risk (VaR), conditional value-at-risk (CVaR), and loss aversion (LA). We find that all of these criteria lead to similar qualitative behaviors regarding how risk-/loss-averse decisions deviate from risk-neutral ones. A particularly interesting result is that the form of demand randomness, i.e., additive or multiplicative, can completely reverse the effect of risk/loss aversion in certain cases. We also find that the risk aversion models, i.e., EUT, MV, MS, VaR, and CVaR, are essentially equivalent in inducing the same set of optimal solutions under adverse market states. However, loss aversion is a mild version of risk aversion in that the LA model can only lead to a subset of the solutions generated by other models. These results may not be valid for other applications, such as the newsvendor pricing problem, so we can conclude that the relationship between alternative risk criteria is case-sensitive and should be evaluated more carefully.

Pricing has been broadly used as an effective tool for companies to control demand. In most operations management models, the demand quantity of a product is formulated as a function of price, which garners the company flexibility to obtain tradeoff between revenue and cost. Such a price-demand function may not be suitable for make-to-order products with heterogeneous demands. To address such an issue, we study the price quotation and scheduling problem for a set of order inquiries. Specifically, a manufacturer receives multiple order inquiries from customers, and needs to quote a price for each inquiry. The quoted price will affect the probability of which the customer will accept the price and confirm the order. Then the manufacturer determines a schedule for processing the confirmed orders. The objective of the manufacturer is to maximize the expected profit, i.e., the total revenue from all confirmed orders less the delay penalty incurred by some orders. We show that the problem is NP-hard, design a heuristic to determine price quotes, and propose a method to find an upper bound of the optimal objective value. The efficiency of our heuristic is evaluated computationally. In general, our heuristic performs weaker with tighter due date, greater scheduling cost relative to order price, and nonlinear relationship between price quote and order placement probability.

By virtue of its cost advantage, online shopping has attracted a substantial population of consumers. At the same time, lacking the first-hand touching experience in online shopping often causes a utility mismatch to consumers, which may not only turn down some potential consumers but also cause a high return rate. A recent business practice for retailers is to encourage showrooming - consumers browsing product in the offline channel but switching to the online channel for purchases, with an aim to leverage both the low cost from the online channel and the real experience from the offline channel. We address such a dual-channel system withomni channels, i.e., concurrently running a mix of online and offline channels by a joint decision on pricing for the two channels as well as the associated return policy. Using an appropriate model we show that a well-designed dual-channel system can indeed increase the profitability of a retailer. Specifically, we find that channel cost difference is the key factor behind the rationale of encouraging showrooming. Retailers can intentionally create a channel price gap to facilitate demand shifting from offline channel to online channel, thus achieving considerable fulfilment cost savings as well as return cost savings. In addition, we reveal that return policy decision is closely related with pricing decisions, and show that return policy design can be viewed as a tool for market segmentation through modulating channel prices. Finally, we identify several barriers for retailers to engage with consumers directly such as consumer valuation uncertainty and cost efficiency of handling returns, and show that it may be beneficial for retailers to engage with consumers indirectly via consumer showrooming.

This paper introduces a new concept, soft precedence constraint (SPC), in machine scheduling problems. Similar to the conventional precedence constraint, SPC specifies some partial order over the jobs; however, an SPC can be violated, but with a certain penalty or cost. The scheduling problem is to balance the tradeoff between the SPC violation penalty and other criteria relative to job completion times. We focus on studying a special case where SPC is defined by a bipartite network. This case is motivated by the berth allocation problem at a transshipment port, where the SPC models any missed container connections from feeder vessels to ocean-going vessels. We discuss the complexity of the problems for different scenarios and develop approximation algorithms. (C) 2019 Elsevier B.V. All rights reserved.

Problem definition: Putting customer experience at the heart of service design has become a governing principle of today’s “experience economy.” Echoing this principle, our paper addresses a service designer’s problem of how to select and sequence activities in designing a service package. Academic/practical relevance: Empirical literature shows an ideal sequence often entails an interior peak; that is, the peak (i.e., highest-utility) activity is placed neither at the beginning nor the end of the package. Theoretic literature, by contrast, advocates placing the peak activity either at the beginning or at the end. Our paper bridges this gap by developing a theory accounting for interior peaks. It also provides managerial implications for activity sequencing and selection. Methodology: We model the activity sequencing and selection problem as a nonlinear optimization problem and reformulate its objective as an additive function to generate structural insights. Results: We show that heterogeneity in memory decay explains the phenomenon of interior peaks. The optimal sequence is in either an “IU” or “UI” shape. An interior peak is optimal when the memory decay rate of the peak activity is neither too high nor too low. Managerial implications: Our research sheds light on service sequencing by weighing the phenomenon of interior peaks. In the presence of an interior peak, we show it is optimal to schedule a low point immediately before or after the peak activity, creating a contrast in customer experience. In addition, interior peaks arise partly because the peak activity is more memorable than others. Guided by this logic, as the peak activity becomes even more memorable, one might be tempted to move it to an earlier slot; we show that, counterintuitively, moving it to a later slot can be optimal. Our research also provides implications for activity selection by showing the optimal portfolio may consist of activities with the highest- and lowest-utility values but not those with medium values.

In this paper we propose a new instrument, a simultaneous penalization and subsidization, for stabilizing the grand coalition and enabling cooperation among all players of an unbalanced cooperative game. The basic idea is to charge a penalty z from players who leave the grand coalition, and at the same time provide a subsidy ω to players who stay in the grand coalition. To formalize this idea, we establish a penalty-subsidy function ωz based on a linear programming model, which allows a decision maker to quantify the trade-off between the levels of penalty and subsidy. By studying function ωz, we identify certain properties of the trade-off. To implement the new instrument, we design two algorithms to construct function ωz and its approximation. Both algorithms rely on solving the value of ωz for any given z, for which we propose two effective solution approaches. We apply the new instrument to a class of machine scheduling games, showing its wide applicability.The e-companion is available at https://doi.org/10.1287/opre.2018.1723.

Talent outsourcing, as a form of sharing economy, has gained growing popularity with numerous innovative marketplaces. In this paper we study a stochastic assignment problem pertaining to talent crowdsourcing motived by designer crowdsourcing, which has not been studied or even remains untapped in the OM literature. In the context of talent crowdsourcing, a platform has a pool of registered talents (e.g., designers) who provide customized intelligent service (e.g., design) for its clients. Due to the subjectivity in evaluating an intelligent service, the platform needs to present to a client multiple designs, though only one, or even none, will be selected at the end. This leads to a challenging resources allocation problem faced by the crowdsourcing platform. We tackle the problem via a stochastic sequential assignment model where a crowdsourcing platform assigns available talent to a stochastic project process.We show the optimal policy follows a staircase functional structure, and provide a sufficient condition under which more designers need to be assigned for the current project with a higher value. Furthermore, we show that the impact of the market volatility might be in an opposite way, determined by the structure of the reward function. Some algorithms have been developed to compute the optimal policy, which can be implemented for the pertaining automatization purpose, e.g., smart contract. Finally, an extensive computational study has been performed with rich and useful managerial insights. For example, it is shown that the developed strategic decision of stochastic sequential assignment can improve the expected reward or expected amount of successful projects while the market volatility becomes large, especially facing sparse resources. In addition, the merging and polling effect becomes fairly notable when the two sub-markets have a heterogenous resource-demand structure.

We study a rescheduling problem faced by multiple job owners sharing a single machine, where jobs need to be rescheduled when the machine becomes unavailable for a period of time. The disruption caused by any new schedule is restricted such that the difference between the completion times in the initial and the new schedules of any job is no more than a given threshold. A natural way to reschedule is to process the jobs in the initial sequence, each as early as possible. This defines a feasible schedule over which cost saving can potentially be achieved by optimal rescheduling as long as the cost saving can be fairly shared by job owners. We define a cooperative game for job owners accordingly, to share the cost saving. Given that the optimization problem is computationally intractable, we find several optimal properties and develop an optimal pseudopolynomial time dynamic programming algorithm for rescheduling. We provide a simple closed form core allocation of the total cost saving for all the jobs, and also provide the Shapley value of the game in a computable form. Then we computationally evaluate the extra cost caused by machine unavailability, the cost saving from optimization relative to the naturally constructed schedule, the likelihood for the Shapley value to be a core allocation, and how the Shapley value allocates cost saving among job owners. Managerial insights are derived from the computational studies. This work contributes to the literature by explicitly incorporating two classic scheduling topics: sequencing game and rescheduling. (C) 2017 Elsevier B.V. All rights reserved.

Piracy attack is a serious safety problem for maritime transport worldwide. Whilst various strategic actions can be taken, such as rerouting vessels and strengthening navy patrols, this still cannot completely eliminate the possibility of a piracy attack. It is therefore important for a commercial vessel to be equipped with operational solutions in case of piracy attacks. In particular, the choice of a direction for rapidly fleeing is a critical decision for the vessel. In this article, we formulate such a problem as a nonlinear optimal control problem. We consider various policies, such as maintaining a straight direction or making turns, develop algorithms to optimize the policies, and derive conditions under which these policies are effective and safe. Our work can be used as a real‐time decision making tool that enables a vessel master to evaluate different scenarios and quickly make decisions.

Motivated by the practices of product design and production outsourcing, we develop a game-theoretic model of a dyadic supply chain to study the joint decisions on product line design and outsourcing. We derive equilibrium decisions on product quality levels and identify the conditions under which the manufacturer outsources production. We first study the case of symmetric outsourcing where the manufacturer outsources both high-end and low-end products, and obtain the following managerial insights. Symmetric outsourcing tends to enlarge product differentiation within a product line, quality levels under symmetric outsourcing are lower than those under symmetric insourcing when the manufacturer's marginal quality cost is medium, and the size of the product line decreases with both the subcontractor's production cost under outsourcing and the manufacturer's production cost under insourcing. When allowing asymmetric sourcing where the manufacturer outsources one product and insources another product, we find that the sourcing strategy reverses the effect of the manufacturer's nonquality cost on quality levels where nonquality cost represents marginal production cost independent of quality level, and asymmetric sourcing may be taken when the manufacturer's nonquality cost is low and symmetric outsourcing should be taken when the manufacturer's nonquality cost is high.

We consider a two-stage supply chain with one supplier and one manufacturer. The manufacturer faces a Poisson demand process where the arrival rate depends on the selling price, the announced delivery time, and the delivery reliability defined as the probability of satisfying the announced delivery time. Such a demand model generalizes the works in the literature by simultaneously considering the above three demand sensitivity factors. The main purpose of this paper is to study the equilibrium decisions in the supply chain with an all-unit quantity discount contract. We consider four scenarios regarding whether the leadtime standard, the delivery reliability standard, and the manufacturer's capacity are endogenous, and whether the manufacturer's production cost is its private information. We find that an all-unit quantity discount scheme can coordinate the supply chain for most cases. Managerial insights are observed regarding the impact of the three demand sensitivity factors. For example, the breakpoint in an optimal quantity discount contract always increases with the delivery reliability sensitivity under an exogenous delivery reliability, but may decrease under an endogenous delivery reliability; with asymmetric information, a higher variance of the manufacturer's unit production costs leads to a lower unit wholesale price for the low-cost manufacturer.